CtrlK
BlogDocsLog inGet started
Tessl Logo

matthew-a-carr/write-adr

Write a new Architecture Decision Record following CONSTITUTION §7, wire it into the `docs/decisions/README.md` index, and update the status lines of any ADR it supersedes. Use when a change meets an ADR trigger in AGENTS.md ("When to write an ADR"), when a human says "write an ADR for X", or when called by `implement-spec` / `draft-spec` at the point a decision is made. Edits files — it creates the ADR and the index row in the same change as the code it documents.

90

1.06x
Quality

97%

Does it follow best practices?

Impact

100%

1.06x

Average score across 2 eval scenarios

SecuritybySnyk

Passed

No known issues

Overview
Quality
Evals
Security
Files

Quality

Content

92%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

This is a well-crafted, highly actionable skill that clearly sequences a multi-step document creation workflow with appropriate validation checkpoints and error handling. It is concise, assumes Claude's competence, and provides concrete patterns (filenames, commit messages, index rows) rather than abstract guidance. The only minor weakness is that the template is inlined despite referencing an existing template file, and the external references (AGENTS.md, CONSTITUTION) cannot be verified without bundle files.

DimensionReasoningScore

Conciseness

Every section earns its place. The skill assumes Claude knows what an ADR is, how markdown works, and how git commits function. The 'Do not' section is tight and addresses real failure modes rather than restating obvious things. No wasted tokens explaining basic concepts.

3 / 3

Actionability

Each step provides concrete, specific instructions: exact filename patterns (NNN-short-descriptive-title.md), exact markdown template with field-by-field guidance, exact commit message format (docs(adr-NNN): <decision>), exact index row format. The triggers list is specific and enumerated, and the 'do not' triggers are equally concrete.

3 / 3

Workflow Clarity

The workflow is clearly sequenced as Steps 0-6 with explicit validation at Step 0 (confirm ADR is warranted — stop if not), collision handling in Step 1 (renumber if number exists on main), supersession handling in Step 4 with bidirectional updates, and a cross-reference sweep in Step 5. The 'Do not' section serves as a checklist against common errors, particularly the missing index row and stale status lines.

3 / 3

Progressive Disclosure

The skill references external files (AGENTS.md, CONSTITUTION §7/§9, docs/decisions/000-template.md, docs/decisions/README.md) clearly and at one level deep, which is good. However, no bundle files are provided, so we can't verify these references resolve. The skill also inlines the full ADR template rather than simply pointing to the template file it already references, which is slightly redundant but arguably justified for clarity.

2 / 3

Total

11

/

12

Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

This is an excellent skill description that clearly articulates what the skill does (creates ADRs, updates the index, updates superseded ADR statuses) and when to use it (explicit trigger conditions from AGENTS.md, user requests, and inter-skill calls). It uses third person voice throughout and provides specific file paths and cross-references that make it highly distinctive and actionable.

DimensionReasoningScore

Specificity

Lists multiple specific concrete actions: write a new ADR following a specific constitution section, wire it into a specific index file, and update status lines of superseded ADRs. Very concrete and actionable.

3 / 3

Completeness

Clearly answers both 'what' (write ADR, update index, update superseded ADRs) and 'when' with explicit triggers ('when a change meets an ADR trigger in AGENTS.md', 'when a human says write an ADR for X', 'when called by implement-spec / draft-spec'). The 'Use when' clause is explicit and comprehensive.

3 / 3

Trigger Term Quality

Includes strong natural trigger terms: 'ADR', 'Architecture Decision Record', 'write an ADR for X', and references to calling skills like 'implement-spec' / 'draft-spec'. Also mentions the AGENTS.md trigger conditions. These are terms users would naturally use.

3 / 3

Distinctiveness Conflict Risk

Highly distinctive — focuses specifically on Architecture Decision Records, references specific files (docs/decisions/README.md, CONSTITUTION §7), and names specific calling skills. Very unlikely to conflict with other skills.

3 / 3

Total

12

/

12

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation11 / 11 Passed

Validation for skill structure

No warnings or errors.

Reviewed

Table of Contents